The search works forwards in time. At each step it asks: from every place the boat could be now, where could it be a little later, sailing at every heading the boat can hold? The furthest points reached at each step form one band. The route is the path back through them.
The bands
Each band is one time step. Their shape tells you more than the track does.
Wide, evenly spaced bands mean the boat is fast and the choice of heading does not matter much. Reaching in steady breeze looks like this.
Bands squeezed together mean slow going: light air, foul current, or beating.
A band with a pronounced lobe means one direction is much better than its neighbours. That is a routing decision being made in front of you, and it is worth understanding before you trust it.
Bands that stop spreading mean the search hit a constraint: land, shallow water, or the edge of the forecast corridor.
Spotting a bad route
The track hugs the edge of the weather. The search wanted to go further out than the forecast it was given covers. Widen the course or lengthen the duration and compute again.
The bands stop well short of the finish. The search ran out of time or forecast. Increase the duration.
The track oscillates in small zigzags. Check the result's notes to see whether the search was simplified for a long passage. The result states when that happened.
A sharp corner in open water with no wind change. Usually a depth constraint. Check the draft on the boat, and remember the depth data is generalised.
Comparing models
The bands you are looking at belong to whichever model the Model picker at the top is showing. Left on Consensus it is the one the others broadly agree with.
Step through the individual models and watch what the bands do. If the lobes point the same way in all of them, the decision is settled. If one model bulges toward the opposite side of the course, that is the race's real question, and you now know about it before the start rather than at the first shift.